Storm_data_visualization
Fiona Fang
2024-03-03
Preparation
Load necessary packages
pacman::p_load(tidyverse, maps, dplyr, ggplot2, ggthemes, sf, urbnmapr, statebins,
data.table, leaflet, RColorBrewer, geosphere, RANN, readr, scales,
geosphere, purrr, htmlwidgets)Load the weather data
storm_data <- read.csv("/Users/jiaying/Desktop/5063_Data_Visualization/07_severe_weather_GRADED/data/storms.csv")
head(storm_data)## BEGIN_YEARMONTH BEGIN_DAY BEGIN_TIME END_YEARMONTH END_DAY END_TIME
## 1 201704 6 1509 201704 6 1509
## 2 201704 6 930 201704 6 940
## 3 201704 5 1749 201704 5 1753
## 4 201704 16 1759 201704 16 1900
## 5 201704 15 1550 201704 15 1550
## 6 201704 29 915 201704 29 1115
## EPISODE_ID EVENT_ID STATE STATE_FIPS YEAR MONTH_NAME EVENT_TYPE
## 1 113355 678791 NEW JERSEY 34 2017 April Thunderstorm Wind
## 2 113459 679228 FLORIDA 12 2017 April Tornado
## 3 113448 679268 OHIO 39 2017 April Thunderstorm Wind
## 4 113697 682042 OHIO 39 2017 April Flood
## 5 113683 682062 NEBRASKA 31 2017 April Hail
## 6 114718 688082 INDIANA 18 2017 April Flash Flood
## CZ_TYPE CZ_FIPS CZ_NAME WFO BEGIN_DATE_TIME CZ_TIMEZONE
## 1 C 15 GLOUCESTER PHI 06-APR-17 15:09:00 EST-5
## 2 C 71 LEE TBW 06-APR-17 09:30:00 EST-5
## 3 C 57 GREENE ILN 05-APR-17 17:49:00 EST-5
## 4 C 25 CLERMONT ILN 16-APR-17 17:59:00 EST-5
## 5 C 25 CASS OAX 15-APR-17 15:50:00 CST-6
## 6 C 155 SWITZERLAND ILN 29-APR-17 09:15:00 EST-5
## END_DATE_TIME INJURIES_DIRECT INJURIES_INDIRECT DEATHS_DIRECT
## 1 06-APR-17 15:09:00 0 0 0
## 2 06-APR-17 09:40:00 1 0 0
## 3 05-APR-17 17:53:00 0 0 0
## 4 16-APR-17 19:00:00 0 0 0
## 5 15-APR-17 15:50:00 0 0 0
## 6 29-APR-17 11:15:00 0 0 0
## DEATHS_INDIRECT SOURCE MAGNITUDE MAGNITUDE_TYPE FLOOD_CAUSE
## 1 0 Trained Spotter 52.0 EG <NA>
## 2 0 Emergency Manager NA <NA> <NA>
## 3 0 Amateur Radio 50.0 EG <NA>
## 4 0 Public NA <NA> Heavy Rain
## 5 0 Trained Spotter 1.5 <NA> <NA>
## 6 0 Law Enforcement NA <NA> Heavy Rain
## CATEGORY TOR_F_SCALE TOR_LENGTH TOR_WIDTH TOR_OTHER_WFO TOR_OTHER_CZ_STATE
## 1 NA <NA> NA NA <NA> <NA>
## 2 NA EF0 7.43 20 <NA> <NA>
## 3 NA <NA> NA NA <NA> <NA>
## 4 NA <NA> NA NA <NA> <NA>
## 5 NA <NA> NA NA <NA> <NA>
## 6 NA <NA> NA NA <NA> <NA>
## TOR_OTHER_CZ_FIPS TOR_OTHER_CZ_NAME BEGIN_RANGE BEGIN_AZIMUTH BEGIN_LOCATION
## 1 NA <NA> 1 NW FRIES MILLS
## 2 NA <NA> 1 S PUNTA RASSA
## 3 NA <NA> 3 NE FAIRBORN
## 4 NA <NA> 1 NW SUMMERSIDE
## 5 NA <NA> 2 ENE COLE ARPT
## 6 NA <NA> 0 N VEVAY
## END_RANGE END_AZIMUTH END_LOCATION BEGIN_LAT BEGIN_LON END_LAT END_LON
## 1 1 NW FRIES MILLS 39.6600 -75.0800 39.6600 -75.0800
## 2 1 SW FORT MYERS VILLAS 26.5010 -81.9980 26.5339 -81.8836
## 3 3 NE FAIRBORN 39.8500 -83.9900 39.8500 -83.9900
## 4 1 NW SUMMERSIDE 39.1065 -84.2875 39.1061 -84.2874
## 5 2 ENE COLE ARPT 40.9800 -95.8900 40.9800 -95.8900
## 6 0 SW VEVAY 38.7500 -85.0700 38.7465 -85.0766
## DATA_SOURCE DAMAGE_PROPERTY_USD DAMAGE_CROPS_USD
## 1 CSV NA NA
## 2 CSV 110000 0
## 3 CSV 1000 0
## 4 CSV 5000 0
## 5 CSV 0 0
## 6 CSV 10000 0Check the dimension of the dataset
dim(storm_data)## [1] 380137 491. Damage from Storms
a) State Level Choropleth Maps
Provide a static state-level choropleth map of the United States
visualizing where monetary damage is recorded (by using the sum of the
variables DAMAGE_PROPERTY_USD and
DAMAGE_CROPS_USD).
# Examine the variables
summary(storm_data$DAMAGE_PROPERTY_USD)## Min. 1st Qu. Median Mean 3rd Qu. Max. NA's
## 0 0 0 273521 0 950000000 77952summary(storm_data$DAMAGE_CROPS_USD)## Min. 1st Qu. Median Mean 3rd Qu. Max. NA's
## 0 0 0 35925 0 500000000 76244Get U.S states map
# Load and prepare the US states map data, converting state names to uppercase
us.states <- map_data("state") %>%
as_tibble() %>%
rename(state_name = region) %>%
select(-subregion) %>%
mutate(state_name = toupper(state_name))
# Create a table with state names, abbreviations, and centers
statenames <- as_tibble(cbind(state_name = toupper(state.name),
state.abb = state.abb,
state.center.x = state.center$x,
state.center.y = state.center$y)) %>%
mutate(across(c(state.center.x, state.center.y), as.numeric))
# Join the map data with state names, abbreviations, and centers
us.states <- left_join(us.states, statenames, by = "state_name")State-level damage data
# Summarize the storm damage by state name (now both in uppercase)
damage_by_state <- storm_data %>%
group_by(STATE) %>%
summarise(Total_Damage_USD = sum(DAMAGE_PROPERTY_USD, na.rm = TRUE) + sum(DAMAGE_CROPS_USD, na.rm = TRUE)) %>%
rename(state_name = STATE)
# Merge the damage data with the full us.states data, including geographic details
damage_data_merged <- left_join(us.states, damage_by_state, by = "state_name")Plot the map
# Plot the choropleth map with state-level damage data
damage_map <- ggplot(damage_data_merged, aes(x = long, y = lat, group = group)) +
geom_polygon(aes(fill = Total_Damage_USD), color = "white") +
scale_fill_gradient(low = "lightblue", high = "red", name = "Total Damage (USD)") +
geom_text(data = statenames, inherit.aes = FALSE,
aes(label = state.abb, x = state.center.x, y = state.center.y), color = "white") +
theme_void() +
coord_fixed(1.3) +
ggtitle("Total Damage by State")
damage_map
Discussion:
From the plot, we can see that Texas (TX), Louisiana (LA), and Florida (FL) are highlighted with the darkest red, suggesting these states have incurred the most monetary damage according to the data. This could be due to severe weather events affecting these states more significantly during the time frame of the data.
State such as Georgia (GA) is colored with a moderate red, indicating a lesser degree of damage compared to the highest damage states but still significant. Most other states are colored in varying shades of blue, indicating relatively lower monetary damages from severe weather events.
There seems to be a pattern where states in the southeast and coastal regions show more damage, which could correlate with the frequency and intensity of storms, such as hurricanes, in those areas.
b) County Choropleth Maps
Provide a static county-level choropleth map of the United States
visualizing where monetary damage is recorded (by using the sum of the
variables DAMAGE_PROPERTY_USD and
DAMAGE_CROPS_USD).
Get U.S. County map
# Retrieve county-level map data
uscounties_sf <- get_urbn_map("counties", sf = TRUE)
# Remove " County" and lowercase
uscounties_sf <- uscounties_sf %>%
mutate(county_name = tolower(trimws(gsub(" County", "", county_name)))) Calculate the damage data by county
# Filter out county-level events
county_events <- storm_data %>%
filter(CZ_TYPE == "C")
# Aggregate the storm data at the county level
county_damage <- county_events %>%
group_by(CZ_FIPS, CZ_NAME) %>%
summarise(Total_Damage = sum(DAMAGE_PROPERTY_USD, na.rm = TRUE) + sum(DAMAGE_CROPS_USD, na.rm = TRUE), .groups = "drop")
# Standardize county names
county_damage <- county_damage %>%
mutate(CZ_NAME = tolower(trimws(CZ_NAME)))
county_damage$CZ_NAME <- gsub("\\(c\\)", "city", county_damage$CZ_NAME)
county_damage$CZ_NAME <- sub("baltimore city city", "baltimore city", county_damage$CZ_NAME, ignore.case = TRUE)
county_damage$CZ_NAME <- sub("bristol bay", "bristol bay borough", county_damage$CZ_NAME, ignore.case = TRUE)
county_damage$CZ_NAME <- sub("carson city city", "carson city", county_damage$CZ_NAME, ignore.case = TRUE)
county_damage$CZ_NAME <- sub("charles city city", "charles city", county_damage$CZ_NAME, ignore.case = TRUE)
# Merge the aggregated damage data with the county-level map data
damage_county_data_merged <- left_join(county_damage, uscounties_sf, by = c("CZ_NAME" = "county_name"))## old-style crs object detected; please recreate object with a recent sf::st_crs()Create the map view
# Plot the choropleth map with county-level damage data
county_damage_map <- ggplot(damage_county_data_merged) +
geom_sf(aes(fill = Total_Damage, geometry = geometry)) +
scale_fill_gradient(name = "Total Damage (USD)", low = "lightblue", high = "red") +
theme(legend.position = "right") +
ggtitle("Total Damage by County") +
theme(plot.title = element_text(hjust = 0.5, size = 20), legend.title = element_text(size = 10)) +
theme_map()
county_damage_map
Highlight state borders
# Get state boundaries
state_boundaries <- get_urbn_map("states", sf = TRUE)
# Plot the county-level choropleth map
county_damage_map <- ggplot() +
geom_sf(data = damage_county_data_merged, aes(fill = Total_Damage, geometry = geometry)) +
scale_fill_gradient(name = "Total Damage (USD)", low = "lightblue", high = "red") +
theme(legend.position = "right") +
ggtitle("Total Damage by County") +
theme(plot.title = element_text(hjust = 0.5, size = 20), legend.title = element_text(size = 10)) +
theme_map() +
geom_sf(data = state_boundaries, fill = NA, color = "white", size = 10)
county_damage_map## old-style crs object detected; please recreate object with a recent sf::st_crs()
## old-style crs object detected; please recreate object with a recent sf::st_crs()
c) Density Maps
Provide an alternative map, in which you highlight the density of severe events by focusing on the variables of injuries and/or deaths associated with storms.
Discuss briefly which of the two approaches provides a better visual summary of the distribution of the destructive effects of storms.
Summarize total injuries and deaths by state
Include both direct and indirect injuries and deaths
storm_summary <- storm_data %>%
group_by(STATE) %>%
summarise(Total_Injuries = sum(INJURIES_DIRECT, na.rm = TRUE) + sum(INJURIES_INDIRECT, na.rm = TRUE),
Total_Deaths = sum(DEATHS_DIRECT, na.rm = TRUE) + sum(DEATHS_INDIRECT, na.rm = TRUE),
Total_Injuries_Deaths = Total_Injuries + Total_Deaths,
.groups = 'drop')Deal with state
# Predefined list of state names and abbreviations
state_abbreviations <- data.frame(
STATE = c("ALABAMA", "ALASKA", "ARIZONA", "ARKANSAS", "CALIFORNIA", "COLORADO",
"CONNECTICUT", "DELAWARE", "FLORIDA", "GEORGIA", "HAWAII", "IDAHO", "ILLINOIS",
"INDIANA", "IOWA", "KANSAS", "KENTUCKY", "LOUISIANA", "MAINE", "MARYLAND",
"MASSACHUSETTS", "MICHIGAN", "MINNESOTA", "MISSISSIPPI", "MISSOURI", "MONTANA",
"NEBRASKA", "NEVADA", "NEW HAMPSHIRE", "NEW JERSEY", "NEW MEXICO", "NEW YORK",
"NORTH CAROLINA", "NORTH DAKOTA", "OHIO", "OKLAHOMA", "OREGON", "PENNSYLVANIA",
"RHODE ISLAND", "SOUTH CAROLINA", "SOUTH DAKOTA", "TENNESSEE", "TEXAS", "UTAH",
"VERMONT", "VIRGINIA", "WASHINGTON", "WEST VIRGINIA", "WISCONSIN", "WYOMING"),
abbreviation = c("AL", "AK", "AZ", "AR", "CA", "CO", "CT", "DE", "FL", "GA", "HI", "ID", "IL",
"IN", "IA", "KS", "KY", "LA", "ME", "MD", "MA", "MI", "MN", "MS", "MO", "MT",
"NE", "NV", "NH", "NJ", "NM", "NY", "NC", "ND", "OH", "OK", "OR", "PA", "RI",
"SC", "SD", "TN", "TX", "UT", "VT", "VA", "WA", "WV", "WI", "WY")
)
# Convert state names to abbreviations
storm_summary <- merge(storm_summary, state_abbreviations, by.x = "STATE", by.y = "STATE", all.x = TRUE)
# Keep only the rows where a match was found in state_abbreviations
storm_summary <- storm_summary[!is.na(storm_summary$abbreviation),]
storm_summary$state <- storm_summary$abbreviation
storm_summary <- storm_summary[ , !(names(storm_summary) %in% c("abbreviation"))]Plot the data using geom_statebins
# Plot the total injuries and deaths data by state
ggplot(storm_summary, aes(state = state, fill = Total_Injuries_Deaths)) +
geom_statebins() +
scale_fill_gradientn(name = "Total Injuries and Deaths",
colours = c("#eff3ff", "#bdd7e7", "#6baed6", "#3182bd", "#08519c"),
values = scales::rescale(c(1, max(storm_summary$Total_Injuries_Deaths, na.rm = TRUE))),
guide = "colourbar") +
theme_statebins() +
labs(title = "Total Injuries and Deaths by State",
subtitle = "Data represented using state bins")
Discussion:
The first visualization uses a geographical map with states drawn to scale and colored according to the total monetary damage caused by storms. This map provides a geographical context that helps viewers recognize the affected areas and understand the spatial distribution of storm damage across the country. The color gradient gives a clear indication of the severity of damage, with darker colors representing more damage. This type of map is intuitive for viewers to understand and can be particularly effective for showing how the effects of storms are spread across different regions. However, the use of actual geographical shapes may give more visual weight to larger states, even if the impact is less severe compared to smaller states.
The second visualization utilizes a grid layout, known as a “statebin” map, where each state is represented by an equally sized square. This approach avoids the size bias inherent in geographical maps, as each state has equal visual weight regardless of its actual size. The color gradient indicates the total number of injuries and deaths, providing a quick visual summary of the human impact of storms without the distraction of varying state sizes. However, since this map abstracts away the actual geography, it may be less immediately recognizable and might not convey the spatial relationships and proximity between states as effectively as a traditional map.
In summary, if the goal is to show the spatial distribution and highlight areas with the greatest financial damage, the first map is more effective. However, if the objective is to present a clear, unbiased comparison of the impact across all states, the second “statebin” map is better.
2. Location of Severe Events
a) Interactive Map of Severe Weather Events
Create a leaflet map of the United States showing the
location of severe weather events which result in at least one death
(hint: use EVENT_TYPE). Ignore locations that fall outside
the United States. Provide at least three pieces of information on the
incident in a popup.
Filter the data
# Filter severe weather events with at least one death and within the United States
severe_events_us <- storm_data %>%
filter(DEATHS_DIRECT > 0, STATE %in% state_abbreviations$STATE)
# Match state names to state abbreviations
severe_events_us <- severe_events_us %>%
mutate(STATE_ABBR = state_abbreviations$abbreviation[match(STATE, state_abbreviations$STATE)])Create the popup
# Create a popup with at least three pieces of information
popup <- lapply(1:nrow(severe_events_us), function(i) {
paste("<b>Event Type:</b>", severe_events_us$EVENT_TYPE[i],
"<br><b>Location:</b>", severe_events_us$STATE[i],
"<br><b>Date:</b>", severe_events_us$BEGIN_DATE_TIME[i],
"<br><b>Direct Deaths:</b>", severe_events_us$DEATHS_DIRECT[i],
"<br><b>Indirect Deaths:</b>", severe_events_us$DEATHS_INDIRECT[i],
"<br><b>Direct Injuries:</b>", severe_events_us$INJURIES_DIRECT[i],
"<br><b>Indirect Injuries:</b>", severe_events_us$INJURIES_INDIRECT[i])
})Create the leaflet map
# Set the Thunderforest API key
thunderforest_api_key <- "fa59a7c912dd4eedb5a614e3a9720d1a"
# Define the Thunderforest tile URL
thunderforest_tile_url <- paste0("https://{s}.tile.thunderforest.com/{style}/{z}/{x}/{y}{scale}.{format}?apikey=",
thunderforest_api_key)
# Create the leaflet map
leaflet(severe_events_us) %>%
setView(lng = -95.7129, lat = 37.0902, zoom = 4) %>% # Centered on the United States
addTiles(urlTemplate = thunderforest_tile_url,
options = list(style = "cycle", format = "png", scale = "", apikey = thunderforest_api_key)) %>%
addMarkers(~BEGIN_LON, ~BEGIN_LAT, popup = popup)b) Color by Type of Weather Event
Start with the previous map. Now, distinguish the markers of the
weather event locations by EVENT_TYPE, i.e. what kind of
weather event occurred. If there are too many categories, collapse some
categories. Choose an appropriate coloring scheme to map the locations
by type of weather event. Add a legend informing the user about the
color scheme. Also make sure that the information about the type of
weather event is now contained in the popup information. Show this
map.
Check event type
unique_event_types <- unique(severe_events_us$EVENT_TYPE)
print(unique_event_types)## [1] "Flood" "Thunderstorm Wind"
## [3] "Wildfire" "Strong Wind"
## [5] "Rip Current" "Tornado"
## [7] "Flash Flood" "Extreme Cold/Wind Chill"
## [9] "Cold/Wind Chill" "Ice Storm"
## [11] "High Surf" "Sneakerwave"
## [13] "High Wind" "Heavy Rain"
## [15] "Heavy Snow" "Lightning"
## [17] "Heat" "Avalanche"
## [19] "Hurricane" "Excessive Heat"
## [21] "Winter Weather" "Storm Surge/Tide"
## [23] "Tropical Storm" "Lakeshore Flood"
## [25] "Winter Storm" "Debris Flow"
## [27] "Coastal Flood" "Blizzard"
## [29] "Hail" "Dust Storm"
## [31] "Frost/Freeze" "Astronomical Low Tide"
## [33] "Lake-Effect Snow"Deal with event types
# Collapse event type
collapse_event_type <- function(event_type) {
event_mapping <- c(
"Flood" = "Water Related",
"Thunderstorm Wind" = "Wind Related",
"Wildfire" = "Fire Related",
"Strong Wind" = "Wind Related",
"Rip Current" = "Water Related",
"Tornado" = "Severe Storm",
"Flash Flood" = "Water Related",
"Extreme Cold/Wind Chill" = "Temperature Extreme",
"Cold/Wind Chill" = "Temperature Extreme",
"Ice Storm" = "Winter Related",
"High Surf" = "Water Related",
"Sneakerwave" = "Water Related",
"High Wind" = "Wind Related",
"Heavy Rain" = "Water Related",
"Heavy Snow" = "Winter Related",
"Lightning" = "Severe Storm",
"Heat" = "Temperature Extreme",
"Avalanche" = "Winter Related",
"Hurricane" = "Severe Storm",
"Excessive Heat" = "Temperature Extreme",
"Winter Weather" = "Winter Related",
"Storm Surge/Tide" = "Water Related",
"Tropical Storm" = "Severe Storm",
"Lakeshore Flood" = "Water Related",
"Winter Storm" = "Winter Related",
"Debris Flow" = "Water Related",
"Coastal Flood" = "Water Related",
"Blizzard" = "Winter Related",
"Hail" = "Severe Storm",
"Dust Storm" = "Wind Related",
"Frost/Freeze" = "Temperature Extreme",
"Astronomical Low Tide" = "Water Related",
"Lake-Effect Snow" = "Winter Related"
)
collapsed_event <- sapply(event_type, function(x) ifelse(x %in% names(event_mapping), event_mapping[x], "Other"))
return(collapsed_event)
}
# Apply the function to the dataset
severe_events_us$collapsed_event_type <- collapse_event_type(severe_events_us$EVENT_TYPE)Create a new popup
popup2 <- lapply(1:nrow(severe_events_us), function(i) {
paste("<b>Collapsed Type:</b>", severe_events_us$collapsed_event_type[i],
"<br><b>Event Type:</b>", severe_events_us$EVENT_TYPE[i],
"<br><b>Location:</b>", severe_events_us$STATE[i],
"<br><b>Date:</b>", severe_events_us$BEGIN_DATE_TIME[i],
"<br><b>Direct Deaths:</b>", severe_events_us$DEATHS_DIRECT[i],
"<br><b>Indirect Deaths:</b>", severe_events_us$DEATHS_INDIRECT[i],
"<br><b>Direct Injuries:</b>", severe_events_us$INJURIES_DIRECT[i],
"<br><b>Indirect Injuries:</b>", severe_events_us$INJURIES_INDIRECT[i])
})Adjust the points
# Create a color palette for the unique collapsed event types
event_types <- unique(severe_events_us$collapsed_event_type)
pal <- colorFactor(palette = "Set3", domain = event_types)
color_event_type <- pal(severe_events_us$collapsed_event_type)
# Create a new leaflet map
map <- leaflet(severe_events_us) %>%
setView(lng = -95.7129, lat = 37.0902, zoom = 4) %>% # Centered on the US
addTiles() %>% # Default OpenStreetMap tiles
addCircles(~BEGIN_LON, ~BEGIN_LAT, color = ~color_event_type, popup = popup2)
# Add a legend to the map
map %>% addLegend("bottomright", pal = pal, values = ~collapsed_event_type, title = "Event Type")The “Water Related” events appear to be concentrated along the Gulf Coast and the Eastern Seaboard, which is consistent with the areas that are often affected by hurricanes and tropical storms, especially during the hurricane season from June through November. Coastal flooding and storm surges contribute to these events. “Severe Storm” events, which could include tornadoes, are likely to be prevalent in the central and eastern United States.
c) Cluster
Add marker clustering, so that zooming in will reveal the individual locations but the zoomed out map only shows the clusters. Show the map with clusters.
Add marker clustering
# Add marker clusters to the map with the specified popup content and color
map2 <- map %>% addCircleMarkers(
lng = ~BEGIN_LON,
lat = ~BEGIN_LAT,
color = ~color_event_type,
popup = popup2,
clusterOptions = markerClusterOptions()
)
# Add a legend to the map
map2 <- map2 %>% addLegend(
position = "bottomright",
pal = pal,
values = ~collapsed_event_type,
title = "Event Type"
)
map23. Severe Events and Cities
For all severe weather event locations, identify the nearest city among the Top 100 largest cities in the United States (here) and calculate the distance between the weather event location and the storm location.
Provide a scatter plot showing the relationship between weather event impact and city population.
Now also visualize the patterns separately for different type of
weather events (use Event Type but feel free to reduce the
number of categories). What do you find?
Note: We did not explicitly discuss distance calculations in
lecture. Two packages that have functions to calculate distances are
geosphere::distGeo() and RANN:nn2().
# Download and read the dataset of the Top 100 largest cities in the US
top_cities <- read_csv("https://raw.githubusercontent.com/plotly/datasets/master/us-cities-top-1k.csv")## Rows: 1000 Columns: 5
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (2): City, State
## dbl (3): Population, lat, lon
##
## â„ą Use `spec()` to retrieve the full column specification for this data.
## â„ą Specify the column types or set `show_col_types = FALSE` to quiet this message.# Create a list-column in `severe_events_us` with nearest city information
severe_events_us <- severe_events_us %>%
rowwise() %>%
mutate(nearest_city_info = list(
if (is.na(BEGIN_LON) | is.na(BEGIN_LAT)) {
list(city = NA, distance = NA, population = NA)
} else {
distances <- geosphere::distGeo(
matrix(c(BEGIN_LON, BEGIN_LAT), ncol = 2),
matrix(c(top_cities$lon, top_cities$lat), ncol = 2)
)
nearest_city_index <- which.min(distances)
list(
city = top_cities$City[nearest_city_index],
distance = distances[nearest_city_index],
population = top_cities$Population[nearest_city_index]
)
}
)) %>%
ungroup()
# Unpack the nearest city info
severe_events_us <- severe_events_us %>%
mutate(
nearest_city = map_chr(nearest_city_info, 'city'),
distance_to_city = map_dbl(nearest_city_info, 'distance'),
nearest_city_population = map_dbl(nearest_city_info, 'population')
)
# Ensure that population data is correctly joined
severe_events_us <- severe_events_us %>%
left_join(top_cities %>% select(City, Population), by = c("nearest_city" = "City"))# Plot 1: Relationship between weather event impact and log of city population
ggplot(severe_events_us, aes(x = log10(nearest_city_population), y = DEATHS_DIRECT)) +
geom_point() +
theme_minimal() +
labs(
x = "Log of Nearest City Population",
y = "Number of Direct Deaths",
title = "Impact of Weather Events vs Log of Nearest City Population"
)## Warning: Removed 1538 rows containing missing values or values outside the scale range
## (`geom_point()`).
# Plot 2: Relationship by collapsed event types with log of city population
ggplot(severe_events_us, aes(x = log10(nearest_city_population), y = DEATHS_DIRECT, color = collapsed_event_type)) +
geom_point() +
theme_minimal() +
scale_color_viridis_d() +
labs(
x = "Log of Nearest City Population",
y = "Number of Direct Deaths",
title = "Impact of Weather Events by Type vs Log of Nearest City Population",
color = "Event Type"
)## Warning: Removed 1538 rows containing missing values or values outside the scale range
## (`geom_point()`).
Discussion:
Distribution of Direct Deaths: The majority of data points cluster at the lower end of the direct deaths scale, which suggests that most weather events result in few or no direct fatalities. There are a few events with a higher number of direct deaths, which stand out above the general trend. These could be severe events that warrant further investigation to understand the factors contributing to higher fatalities.
Population Relationship: There doesn’t appear to be a strong correlation between the size of the nearest city population (even on a logarithmic scale) and the number of direct deaths. This suggests that the severity of a weather event, in terms of direct fatalities, is not necessarily related to the population size of the nearest city.
Event Types: In the second plot, weather events are categorized by type. The distribution of direct deaths across different event types also shows that most events, regardless of type, result in few or no direct deaths. Severe storms and water-related events seem to have a few cases with higher fatalities. This might be characteristic of these types of events, possibly due to their sudden onset, intensity, or the specific circumstances of the areas they impact.